Transit method is used to detect the presence of exoplanets around distant stars. Here’s a helpful visualization:
Data
Data was obtained from the Kepler space telescope. Data format: time series (equal intervals) with some missing data
Only the light flux was used.
Data is not as clean as expected.
Universe is large - stars exhibit variable brightness at large distances due to several phenomenon.
Not all crests and troughs are relevant.
Components
Results 1: Learning
Results 2: Testing
Scope for Improvement
Hyperparameters such as sequence length, number of hidden units in LSTM layers, number of dense layers, number of neurons in dense layer, etc. can be tuned.
An algorithm can be used to perform phase folding to identify number of planets and period of revolution of each planet around the star. This will also enrich the data.